Commit Graph

1019 Commits

Author SHA1 Message Date
Alexander A. Klimov 1343fd538d Start ApiListener#SyncClient() in the thread pool
... not hosting the coroutines not to block them.

Otherwise a large replay log would block messages sending
until the peer disconnects us.
2020-11-24 17:25:43 +01:00
Alexander Aleksandrovič Klimov 3dcc6c32f3
Merge pull request #8479 from Icinga/bugfix/close-anonymous-connections
Close anonymous connections after 10 seconds
2020-11-24 16:44:09 +01:00
Julian Brost 2a2924855f
Merge pull request #7922 from Icinga/feature/http-status-codes-in-icinga-mainlog-7053
Include HTTP status codes in log
2020-11-24 16:35:58 +01:00
Julian Brost da407660f2
Merge pull request #8500 from Icinga/bugfix/config-sync-only-remove-files-if-timestamp-changed
Config sync: Only remove files, if timestamp changed
2020-11-24 16:34:12 +01:00
Julian Brost c154d4d50e
Merge pull request #8466 from Icinga/feature/one-connection
ApiListener#NewClientHandlerInternal(): reject connections from already connected endpoints
2020-11-24 16:33:15 +01:00
Noah Hilverling 83b4d8e69d Config sync: Only remove files, if timestamp changed 2020-11-24 10:44:38 +01:00
Alexander Aleksandrovič Klimov 39bc1590f6
Merge pull request #8440 from Icinga/bugfix/message-routing-for-global-zones
Fix cluster message routing for global zones
2020-11-24 10:41:17 +01:00
Alexander Aleksandrovič Klimov e84a4a290d
Merge pull request #8450 from Icinga/bugfix/do-not-accept-api-updates-for-unknown-zone
API: Don't accept object updates for unknown global zone
2020-11-24 10:40:20 +01:00
Julian Brost 70c9d49ebc ApiListener: merge new config validation and actication functions
Merge AsyncTryActivateZonesStage and TryActivateZonesStageCallback and
name the result TryActivateZonesStage. The old split was a leftover from
the one being a callback function with no actual meaningful separation.
2020-11-17 09:37:13 +01:00
Noah Hilverling 2d1980c10d
Merge pull request #8476 from Icinga/docs/api-action-api-function
Clarify difference between API actions and functions in their docstrings
2020-11-17 08:17:05 +01:00
Julian Brost e4610e7dbd Use std::mutex instead of Spinlock 2020-11-16 17:38:03 +01:00
Julian Brost 74b65f1642 API filesync: wait for validation process to exit
This avoid having to pass a lock implictly using the captured variables
of a lambda.
2020-11-16 17:10:57 +01:00
Julian Brost d1edcb909c Close anonymous connections after 10 seconds
Anonymous connections are normally only used for requesting a
certificate and are closed after this request is received. However, the
request is only sent if the child has successfully verified the
certificate of its parent so that it is an authenticated connection from
its perspective. In case this verification fails, both ends view it as
an anonymous connection and never actually use it but attempt a
reconnect after 10 seconds leaking the connection. Therefore close it
after a timeout.
2020-11-12 18:01:11 +01:00
Alexander Aleksandrovič Klimov 8ca765d730
Merge pull request #8455 from Icinga/bugfix/replay-object-deletion
Log config object deletions to replay log
2020-11-12 15:08:55 +01:00
Julian Brost 01a278bb5e Clarify difference between API actions and functions in their docstrings 2020-11-12 14:23:41 +01:00
Noah Hilverling 5f6042d92f Fix 'emoving' typo 2020-11-09 16:35:16 +01:00
Julian Brost cb476172ec Fix cluster message routing for global zones
RelayMessageOne used to relay the message only to one other endpoint for
other zones, which is fine, as long as the target zone is a child/parent
zone but breaks if the target zone is a global one. In this case, the
message has to be forwarded within the local zone as well as to one node
in each child zone.
2020-11-09 15:43:43 +01:00
Julian Brost be53b0af9e Log config object deletions to replay log
The initial config object sync for each new connection (in
`ApiListener::SendRuntimeConfigObjects()`) only considers currently
existing objects and has no way to pass the information that objects
were deleted in the meantime.

This commit logs config object deletions to the replay log if required
so that there is a chance that it will be propagated to nodes that were
offline when the deletion happened.

Note that this can only be considered a workaround as the replay log
might be pruned or could even be completely disabled. Also, there still
seems to be a race-condition between the config sync and replay log of
multiple new connections at the same time.
2020-11-09 14:09:44 +01:00
Alexander A. Klimov 29e5d7def7 Include HTTP status codes in log
refs #7053
2020-11-09 10:20:13 +01:00
Noah Hilverling 8ba5f72533 API: Don't accept object updates for unknown zone 2020-11-06 17:27:10 +01:00
Alexander Aleksandrovič Klimov 1450e1bb7f
Merge pull request #8108 from Icinga/bugfix/api-incorrect-response-header-6747
API: Send Content-Type as api response header too
2020-11-03 18:50:31 +01:00
Alexander Aleksandrovič Klimov 939f4591a4
Merge pull request #8087 from Icinga/bugfix/log-cout-permission-error-8086
Display Logmessage if an permission error occurs
2020-11-03 17:23:06 +01:00
Yonas Habteab 488e6bfb67 HTTP Request: Log an exception message if an error occurs 2020-11-02 15:01:48 +01:00
Alexander Aleksandrovič Klimov 1e281b060a
Merge pull request #7952 from Icinga/fix/SO_REUSEPORT-optional
apilistener: Make SO_REUSEPORT optional
2020-10-29 15:56:56 +01:00
Alexander A. Klimov bb851b0558 Merge branch 'master' into feature/v1-actions-execute-command-8034 2020-10-28 18:37:08 +01:00
Alexander A. Klimov 38110e55d3 ApiListener#NewClientHandlerInternal(): reject connections from already connected endpoints 2020-10-26 15:20:58 +01:00
Alexander Aleksandrovič Klimov a32c1bf910
Merge pull request #7864 from Icinga/bugfix/icinga2-doesn-t-close-connections-7203
Add timeout for boost::asio::ssl::stream#async_shutdown()
2020-10-19 15:25:12 +02:00
Alexander A. Klimov 40ac05c182 Introduce Endpoint#capabilities
refs #8034
2020-10-19 13:04:20 +02:00
Alexander Aleksandrovič Klimov f60ae93717
Merge pull request #7870 from Icinga/bugfix/last-zone-sync-stage-validation-failed-7642
Clear ApiListener#last_failed_zones_stage_validation on config::Update if config not changed
2020-10-19 12:31:01 +02:00
Alexander Aleksandrovič Klimov 4bb6f4c910
Merge pull request #8364 from Icinga/bugfix/configsyncstagelock-unlock-owner-213
Make ApiListener::m_ConfigSyncStageLock a SpinLock
2020-10-14 16:01:00 +02:00
Alexander A. Klimov fbfa931b31 Clear ApiListener#last_failed_zones_stage_validation on config::Update if config not changed
refs #7642
2020-10-14 12:17:14 +02:00
Alexander A. Klimov a083635de4 Make ApiListener::m_ConfigSyncStageLock a SpinLock 2020-10-14 10:47:08 +02:00
Julian Brost e04d618ede Catch exceptions in the thread running HandleConfigUpdate
With dc3062a9b0, exceptions in this code
path were no longer caught properly. This commit restores exception
handling for this function.
2020-10-13 14:07:44 +02:00
Yonas Habteab 3f3e04eb69 Allow to create api user w/ password
fixes #8164
2020-10-09 20:15:44 +02:00
mcodato 730075a177
Merge pull request #1 from Al2Klimov/version
Introduce Endpoint#icinga_version
2020-08-05 09:23:28 +02:00
Noah Hilverling a615b2126e
Merge pull request #8142 from Icinga/bugfix/don-not-close-connection-on-missing-heartbeat-8095
Remove all codes related to the heartbeat timeout
2020-07-29 15:33:22 +02:00
Noah Hilverling 97fc70ccb2
Merge pull request #7836 from Icinga/bugfix/jsonrpcconnection-m_seen
Consider a JsonRpcConnection being seen on a single byte of TLS payload, not only a whole message
2020-07-29 15:02:48 +02:00
Yonas Habteab 964a90fa4b Remove all codes related to the heartbeat timeout
until now, if the timeout is exceeded, the connection is immediately terminated.
But since we do not want to disconnect even if the timeout is exceeded, it is
better to send the messages without timeout and have deleted everything that
related to the heartbeat timeout. We also have another mechanism in
JRPC::CheckLiveness that does the disconnect.
2020-07-29 14:27:55 +02:00
Noah Hilverling 9f57e895f1
Merge pull request #8102 from Icinga/bugfix/send-heartbeat-less-often-8098
Send heartbeat every 20s and not 10s
2020-07-21 09:46:33 +02:00
Yonas Habteab df2d30e6af Send Content-Type as api response header too 2020-07-10 10:42:55 +02:00
Yonas Habteab cf5ec5e341 Send heartbeat every 20s and not 10s 2020-07-09 13:22:08 +02:00
Alexander A. Klimov 645dcbdc9e Introduce Endpoint#icinga_version
... and set it to e.g. 21200 via icinga::Hello.
2020-07-07 18:24:06 +02:00
Mattia Codato b07471d803 AuthenticatedApiUser name refactor 2020-07-03 11:17:36 +02:00
Mattia Codato 08c6e9bc42 Make authenticatedApiUser a static property of ActionsHandler 2020-07-03 11:00:40 +02:00
Mattia Codato 27a6fd6b40 Get API user from ActionsHandler 2020-07-03 10:16:23 +02:00
Benjamin Groeber cc6fa66ba6 Allow apiactions to return sucecss http codes different from 200 2020-06-23 11:52:22 +02:00
Alexander A. Klimov 19c632e44b Add timeout for boost::asio::ssl::stream#async_shutdown()
refs #7203
2020-06-17 10:33:35 +02:00
Alexander A. Klimov 647f1547a9 Generalize I/O timeout emulation 2020-06-17 10:31:40 +02:00
Noah Hilverling 84b052b314
Merge pull request #7926 from Icinga/bugfix/jsonrpcconnection-handleandwriteheartbeats-m_endpoint-getname
JsonRpcConnection#HandleAndWriteHeartbeats(): check !!#m_Endpoint
2020-06-03 15:46:38 +02:00
Noah Hilverling d5d89b7f39
Merge pull request #7970 from Icinga/bugfix/reconnect-loop
RequestCertificateHandler(): don't disconnect nodes already integrated into the cluster
2020-04-27 13:05:22 +02:00
Alexander A. Klimov 5a5cf1a2eb RequestCertificateHandler(): don't disconnect nodes already integrated into the cluster
... not to cause a reconnect loop.
2020-04-08 13:29:55 +02:00
Alexander Aleksandrovič Klimov 2e22ceb23e
Merge pull request #7936 from Icinga/bugfix/config-sync-failed-reload-7742
ApiListener::ConfigUpdateHandler(): make the whole process mutually exclusive
2020-04-07 15:55:14 +02:00
Markus Frosch b5fa7569f2 apilistener: Make SO_REUSEPORT optional 2020-03-31 18:03:59 +02:00
Michael Insel 51e534ff4c Fix CA verification regression
Uninitialized bool values may evaluate to true while it should be false.
2020-03-29 16:05:29 +02:00
Alexander A. Klimov 38f3108c1a ApiListener::HandleConfigUpdate(): make the whole process mutually exclusive
refs #7742
2020-03-23 17:33:14 +01:00
Alexander A. Klimov dc3062a9b0 ApiListener::ConfigUpdateHandler(): block as less as possible
refs #7742
2020-03-23 17:31:59 +01:00
Alexander A. Klimov 5e7a675009 JsonRpcConnection#HandleAndWriteHeartbeats(): check !!#m_Endpoint 2020-03-18 11:58:27 +01:00
Noah Hilverling 4c9e4959f3
Merge pull request #7823 from Icinga/bugfix/unify-application-start-times
Fix timing point for Application::GetStartTime() (related to command endpoint grace period)
2020-03-09 09:45:57 +01:00
Noah Hilverling c9ab04d511
Merge pull request #7841 from Icinga/bugfix/jsonrpcconnection-sendmessage-keepalive
JsonRpcConnection#Send*Message(): keep this alive
2020-03-03 10:46:33 +01:00
Michael Friedrich 13d2416e29 Fix regression from JsonRPC PKI CA verification checks
refs #7835
2020-02-27 12:31:02 +01:00
Michael Friedrich 456b0779bb JsonRpcConnection PKI: Document swalled exception 2020-02-20 15:15:54 +01:00
Alexander A. Klimov 0f84ce0470 Consider a JsonRpcConnection being seen on a single byte of TLS payload, not only a whole message 2020-02-19 11:11:53 +01:00
Michael Friedrich a91b9f2ddf Pki: Extend GetCertificateInformation() with version, serial, signature algorithm, SANs 2020-02-17 17:44:10 +01:00
Michael Friedrich 24397fbee8 CA Proxy: Catch exceptions from VerifyCertificate() 2020-02-17 17:43:11 +01:00
Alexander A. Klimov fbce756007 JsonRpcConnection#Send*Message(): keep this alive 2020-02-17 16:12:07 +01:00
Michael Friedrich d53eb34520 Unify Application::GetStartTime() and drop GetMainTime()
This essentially moves the start time into the scope when main
starts to "do something", after the reload and configuration handling
is done.
2020-02-11 17:26:15 +01:00
Michael Friedrich d99f76bdae Merge pull request #7767 from Icinga/bugfix/replay-log-windows-rename
Ensure that log replay files are properly renamed on Windows
2020-02-10 11:42:25 +01:00
Michael Friedrich 8324970ab9
Merge pull request #7809 from Icinga/bugfix/handshake-timeout-7805
Close connections w/o sucessful TLS handshakes during 10s
2020-02-07 16:01:30 +01:00
Michael Friedrich 7488d2614b
Merge pull request #7811 from Icinga/bugfix/config-sync-log
Cluster config sync: Use a more friendly message when configs are equal and don't need a reload
2020-02-06 14:32:46 +01:00
Alexander A. Klimov 35c9f2e7b3 Close connections w/o sucessful TLS handshakes during 10s
refs #7805
2020-02-05 15:53:56 +01:00
Michael Friedrich cdfc6e6f50 Cluster config sync: Use a more friendly message when configs are equal and don't need a reload
This also adds an entry to the troubleshooting docs.
2020-02-05 15:31:41 +01:00
Michael Friedrich 0da46c1d4b Ensure that log replay files are properly renamed on Windows
rename() without _unlink() before doesn't work on Windows.
This commits also adds an error message which was swallowed
previously.
2020-01-20 11:41:43 +01:00
Alexander A. Klimov 3d841d5f64 Don't send event::Heartbeat to unauthenticated peers
refs #7746
2020-01-13 11:21:38 +01:00
Michael Insel 9d55a8264d Fix open connections when agent waits for CA approval
This closes the agent connection when the certificate sign requests
waits for CA approval.

refs #7680
2019-12-03 21:19:39 +01:00
Matthias Baur 177c45d787
Improve error message for POST queries
If the user does a POST query but forgets to add the
'X-HTTP-Method-Override: GET' HTP header, the error message was
misleading. This changes the error message to a more detailed message
which might give the user a better understanding what the problem
could be.

Fixes #7675.
2019-12-03 11:39:40 +01:00
Michael Insel 859658ab98 Fix TLS context copies in ApiListener
This avoids copying the TLS context in the ApiListener class and removes
the obsolete variable.

This is a follow-up from #7654
2019-11-29 16:03:38 +01:00
Michael Insel 8431ea52ee Make SSL context const again
Turns out that the switch to the class member variable inside the lambda
expression is sufficient to fix the issue.
2019-11-25 23:42:32 +01:00
Michael Insel 016a6c3f25 Fix ApiListener const SSL context
This makes the SSL context in ApiListener::SpawnCoroutine non-const to
address an issue when an agent recieves an signed certificate from the
master and tries to update the SSL context. This also uses the class
member variable which is captured by `this` in the lamda expression.

fixes #7650
2019-11-25 22:21:09 +01:00
Michael Friedrich b1787883f7
Merge pull request #7535 from tigercomputing/Icinga/feature/config-stage-activate-parameter
API /v1/config/stages 'activate' parameter
2019-11-15 12:58:03 +01:00
Michael Friedrich 38080405df
Merge pull request #7528 from Icinga/bugfix/api-put-error-handling
API: Handle permission exceptions soon enough, returning 404
2019-11-15 11:53:59 +01:00
Alexander A. Klimov f601ba51e0 Revert "Eventqueue: Remove unused code"
This reverts commit a7873da89d.
2019-11-02 14:00:23 +01:00
Michael Friedrich cb20b4829a Cluster Config Sync: Check the timestamp prior to config file checksums
Otherwise old configuration received from a secondary master/satellite
could always trigger a config change & reload.
2019-10-22 14:07:10 +02:00
Alexander A. Klimov aa4cad7482 Replace std::shared_ptr<Expression> with Expression::Ptr
refs #7361
2019-10-21 17:10:51 +02:00
Alexander A. Klimov ba1ce9c853 Replace std::shared_ptr<boost::asio::ssl::context> with Shared<boost::asio::ssl::context>::Ptr 2019-10-21 16:12:46 +02:00
Alexander A. Klimov a1fef92835 Replace std::shared_ptr<boost::asio::ip::tcp::acceptor> with Shared<boost::asio::ip::tcp::acceptor>::Ptr 2019-10-21 16:12:46 +02:00
Alexander A. Klimov a1683568a1 Replace std::shared_ptr<AsioTlsStream> with Shared<AsioTlsStream>::Ptr 2019-10-21 16:12:35 +02:00
Michael Friedrich e7c07062f8 Avoid syncing .authorititative marker received from < 2.11 parent nodes 2019-10-02 10:49:02 +02:00
Chris Boot aece8d61e2 Introduce 'activate' parameter to config stage handling
This functionality allows a stage to be uploaded and validated as
normal, but not activated. This is useful to pre-flight an Icinga config
package before it is applied to a monitoring cluster, for example in a
CI pipeline.
2019-09-24 17:17:19 +01:00
Michael Friedrich f419efd778 API: Handle permission exceptions soon enough, returning 404
fixes #7513
2019-09-23 09:48:50 +02:00
Michael Friedrich eddb40a913 CSR Auto-signing: Add debug logging for skipped signing 2019-09-18 11:53:58 +02:00
Michael Friedrich d3eb62301e API: Add AES128-GCM-SHA256 compatible cipher for el6
fixes #7501
2019-09-16 14:19:22 +02:00
Michael Friedrich 9ec246a2f4 Revert: Always reset Boost beast buffer in HttpServerConnection#ProcessMessages #7476
Not a simple revert but also adds a comment for the buffer.

refs #7476
2019-09-12 17:00:17 +02:00
Michael Friedrich 2c0e0da2d9 Introduce IoEngine::SpawnCoroutine wrapping asio::spawn and Boost exceptions
This is required to

- catch all exceptions and wrap them into Boost exceptions. They
are the only ones allowed with Boost.Coroutine.
- set a dedicated coroutine stack size for Windows.

refs #7431
2019-09-09 16:40:35 +02:00
Michael Friedrich 5fa7331cc9 Quality: Replace deprecated Boost IO service code
https://github.com/boostorg/asio/issues/110
https://www.boost.org/doc/libs/1_66_0/doc/html/boost_asio/example/cpp03/services/logger_service.hpp
2019-09-09 15:27:57 +02:00
Alexander A. Klimov 0a9563b3db HttpServerConnection#ProcessMessages(): avoid I/O if shutting down
refs #7431
2019-09-09 13:37:44 +02:00
Alexander A. Klimov b85b8b9697 HttpServerConnection#ProcessMessages(): avoid I/O after boost::asio::error::operation_aborted
refs #7431
2019-09-09 13:29:47 +02:00
Alexander A. Klimov dfaeb88ac3 {HttpServer,JsonRpc}Connection#Disconnect(): cancel I/O ops ASAP
refs #7431
2019-09-09 13:11:51 +02:00
Michael Friedrich b3c48e7520
Merge pull request #7485 from Icinga/bugfix/api-disconnect-defer
Avoid the Defer-Disconnect destructor pattern with Boost.Coroutines
2019-09-09 11:41:34 +02:00
Michael Friedrich c5aa978912 Rewrite error handling in HttpServerConnection#EnsureValidHeaders()
Throwing local exceptions unnecessarily pollutes the exception
stack with immediate unwinding. Avoid this pattern at all cost within
Boost.Coroutines. MSVC may handle exceptions differently and cause
problems with stack unwinding.

refs #7431
refs #7351
2019-09-09 11:03:21 +02:00
Michael Friedrich 17d4d17307 Avoid the Defer-Disconnect destructor pattern with Boost.Coroutines
Exceptions in Disconnect() might be thrown (this has been reworked
into error_code locally) which are swallowed inside the Destructor
for being dangerous. On the other hand, swallowing them may
corrupt the stack unwinding operation from the coroutine layer.

The best is to avoid Defer inside lib/remote and call Disconnect()
directly after breaking from other operations.

refs #7351
refs #7431
2019-09-09 10:57:13 +02:00